The Concept of Potential Competition - OECD
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
The Concept of Potential Competition
1
Concept of potential competition
ABUSE OF DOMINANCE IN DIGITAL MARKETS © OECD 2020
PUBE2
Please cite this paper as:
OECD (2021), Concept of potential competition,OECD Competition Committee
Discussion Paper, http://oe.cd/tcpc
This work is published under the responsibility of the Secretary-General of the OECD. The opinions
expressed and arguments employed herein do not necessarily reflect the official views of the OECD or of
the governments of its member countries or those of the European Union.
This document and any map included herein are without prejudice to the status or sovereignty over any
territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city,
or area.
© OECD 2021
CONCEPT OF POTENTIAL COMPETITION © OECD 20213 Foreword This paper discusses the concept of potential competition as an important pro-competitive factor. While potential competition is inevitably subject to significant uncertainty, where it does exist, the paper suggests treating potential competition with a parity of esteem with respect to actual competition. The paper considers the benefits of extending the timeframe used to evaluate potential competition and reviews the tools that are available to assess it. It suggests such tools may be helpfully placed within a specific framework to enable assessment under the different and greater uncertainty that exists over potential competitive constraints. These tools include many that are already widely used, such as the additional weight placed on credible contemporaneous internal documents, progress against regulatory checkpoints, understanding of business models and of competition to innovate. Similarly, on the counterfactual it suggests following existing best practices such as pro-actively exploring alternative counterfactuals. Other suggestions involve the use of what in some jurisdictions might be newer tools – valuation analysis, forward-looking consumer surveys, spillover analysis of non-overlapping products in adjacent markets, and the development of specialist progress-to-market expertise. The paper also highlights existing trends by competition agencies to advocate for a change in existing decision-making frameworks to effectively protect against the loss of potential competition. In this respect, the paper suggests that there might be a case for using different thresholds for potential competition from those that are used when the concern is over the possible loss of an actual constraint. This paper was written by Chris Pike and Takuya Ohno of the OECD Competition Division, with comments from Antonio Capobianco, Renato Ferrandi, Ruben Maximiano, Wouter Meester and Pedro Caro de Sousa from the OECD Competition Division. It was prepared as a background note to discussions on the concept of potential competition at the 135th meeting of the OECD Competition Committee on 10 June 2021, https://www.oecd.org/daf/competition/the-concept-of-potential-competition.htm CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
4 DAF/COMP(2021)3
CONCEPT OF POTENTIAL COMPETITION @ OECD 20215 Table of contents 1. Introduction 7 2. What is Potential Competition? 9 2.1. Definitions 9 2.2. Which types of markets feature potential competition? 9 2.3. Theories of harm to potential competition 10 2.4. The key parameters to identify potential competition 13 3. Assessing Barriers to Entry 14 3.1. Definitions and forms of barriers to entry 14 3.2. What impact do barriers to entry have on potential competition? 16 4. Assessing the Likelihood and Strength of Potential Competition 20 4.1. How likely and how strong do the thresholds for potential constraint need to be? 20 4.2. How do we assess likelihood and strength of a potential constraint? 24 4.3. Ex-post assessment of potential competition 31 4.4. Counterfactual analysis in potential competition cases 32 5. Assessing the Timeframe in which Potential Competition will emerge 34 5.1. Problems with a short timeframe 35 5.2. How do we assess the timeframe of entry? 36 6. Conclusion 39 References 41 Boxes Box 1. Pacific Biosciences / Illumina - merger and monopolisation case 11 Box 2. Visa/Plaid – merger and monopolisation case 11 Box 3. What to do about cross-platform network effects and behavioural biases? 15 Box 4. Sabre/Farelogix merger 21 Box 5. GSK patent settlement agreement case in Korea 23 Box 6. Dow/DuPont merger 26 Box 7. Cornershop/Uber - Analysis of potential competition by the Chile competition authority 28 Box 8. Do’s and don’ts when assessing potential competition 31 Box 9. Do’s and don’ts for the counterfactual analysis in potential competition cases 33 Box 10. Analysis of potential competition in Siemens/Alstom 37 Box 11. TPG/Vodafone - Analysis of potential competition in Australia 38 CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
6 DAF/COMP(2021)3
CONCEPT OF POTENTIAL COMPETITION @ OECD 20217 1 Introduction Scholars of the Economic Anti-Monopoly movement have recently argued to the US Congress that antitrust laws, as interpreted and enforced today, are inconsistent with modern economic thinking and inadequate to confront and deter growing market power (Baker et al., 2020[1]). They call for the strengthening of antitrust laws, and, in particular, for clarification that the antitrust laws protect potential competition. They also call for legislation that would remove existing precedent established by the courts that limit the scope for antitrust action for the establishment of legal rules that, in appropriate cases, require defendants to prove their conduct does not harm competition, and for increased penalties and enforcement resources.1 In the same hearings before the US Congress where the Economic Anti-Monopoly movement made its proposals, the US Congress was told by a number of other prominent economists and lawyers that there exists a broad consensus about the analytical framework and economic toolkit for evaluating whether a merger is likely to substantially lessen competition, including with respect to nascent and potential competition in digital markets (Barnett et al., 2020[2]). The question that this debate raises is whether antitrust laws are inconsistent with modern economic thinking on potential competition. It would appear that if there is a ‘broad consensus’ on the modern economic thinking on potential competition, as (Barnett et al., 2020[2]) suggest, then many leading economists do not consider that this consensus is adequately reflected in current antitrust law on potential competition. The US is not the only jurisdiction where there is a lively debate on the importance of potential competition. In Australia, for instance, the ACCC has already recommended changes to Australia’s merger laws to expressly require consideration of the effects of a merger on potential competition. In Japan, the merger guidelines and notification thresholds have been revised to ensure that the JFTC can identify mergers or conduct that damages potential competition (JFTC, 2019[3]). In Europe, there is also widespread concern at the elimination of potential competition, for example through acquisitions and exclusionary behaviour in digital and pharmaceutical markets (OECD, 2020[4]). For example in the UK, the CMA’s chief economist identified that it is now uncontroversial to say that there has been a record of under-enforcement in relation to the loss of potential competition (Walker, 2020[5]). Indeed, the CMA has said that it has started to adapt its theories of harm in recent cases to take greater account of the impact on potential competition (CMA, 2019[6]). Meanwhile in Germany and Austria, merger notification processes have been amended to allow both agencies to investigate acquisitions of nascent potential competitors. However, there has also been debate in Europe over the treatment of entry by potential competitors which might mitigate otherwise anti-competitive mergers or conduct. In particular, this has become a fiercely contested issue in cases that sit at the intersection of competition and industrial policy. For example, the Siemens/Alstom merger (see Box 10) that would have formed a European rail-making ‘champion’ featured precisely such an assessment of the threat from potential competitors in China. CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
8 DAF/COMP(2021)3
All the considerations above open up the debate on the concept of potential competition itself, its limits, its
relationship with barriers to entry, how we assess the likelihood, strength and timing of potential constrains,
and the thresholds that we use to make our decisions. This paper therefore explores each of these while
linking the issues to relevant merger, collusion and exclusionary cases in which they have arisen in practice.
The structure of the paper is as follows:
Section 2 discusses the notion of potential competition and theories of harm to potential
competition and introduces key parameters to identify it.
Section 3 explores barriers to entry that can influence the strength of potential competition and
suggests that an assessment of barriers to entry might not provide as such a clear answer to
potential competition in cases where a specific potential competitive constraint is allegedly lost.
Section 4 investigates how to assess the likelihood and strength of a potential competitive
constraint. It also discusses how likely and strong a potential competitive constraint needs to be
for it to be relevant for a decision.
Section 5 discusses the timeframe in which a potential competitive constraint needs to manifest.
Section 6 concludes.
CONCEPT OF POTENTIAL COMPETITION @ OECD 20219 2 What is Potential Competition? Definitions Potential competition could be defined as a competitive constraint on a firm’s behaviour that might potentially arise, but has not yet actually done so. Potential competitive constraints may therefore usefully be distinguished from a potential competitor or entrant who might already be imposing an existing competitive constraint on a firm’s behaviour, despite itself not yet competing on that market. For instance, a potential competitor may currently be selling products that do not compete, or it may not be selling at all. In either case, if the possibility that it will enter is already affecting a firm’s behaviour, then the constraint is an existing one. Focussing on existing constraints, whether from existing rivals or potential entrants, should be much easier since their impact should be observable within existing market data and internal documents (that is the market should already have priced these in). These existing constraints have in the past been confusingly referred to as ‘perceived potential competition’.2 This refers to the fact that the competitive constraint is already perceived by the incumbent, even if entry is still ‘potential’ since it is yet to occur. These constraints have been contrasted with the equally confusingly named, ‘actual potential competition’. These are not yet constraining the incumbent’s behaviour but are expected to do so. (Werden & Limarzi, 2010[7]) suggest that for 50 years the terms ‘actual potential’ and ‘perceived potential’ competition have fostered mistaken notions concerning the assessment of competitive effects. In particular, they point out that labelling the eliminated competition as potential is unhelpful and misleading when that competition already exists. For clarity, we will therefore refer to ‘actual potential competition’ as simply potential competition. In contrast, we will refer to ‘perceived potential competition’ as actual competition. Which types of markets feature potential competition? Potential competitive constraints are likely to be important in many markets across the economy, and can be subject to antitrust analysis. Areas of recent focus have been acquisitions of start-ups and pay-for-delay agreements in pharmaceutical and biotechnology markets (Cunningham, Ederer and Song, 2018[8]) (Colino et al., 2017[9]). In these markets, actual competition is often restricted by intellectual property rights, and potential competition can be observed and assessed using the regulatory pipeline of products going through trials with medical regulators. Similar processes apply to medical technology markets and agriculture markets (see for example Dow/Du Pont at Box 6). The development of new pipelines or other infrastructure which require planning or other regulatory permission also have highly structured procedures preceding access to market that allow for the identification of potential competitors. CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
10 DAF/COMP(2021)3
The debate on the importance of potential competition and how to assess it under competition law,
however, has also extended to unregulated markets in which innovation is important, for example in digital
markets. These include multi-sided platform markets where large network effects can tip markets towards
a single firm, meaning that there are few actual constraints, and hence any remaining potential constraint
is particularly important (LEAR, 2019[10]). Potential competition constraints are also important for any
bidding market, for example for the award of concessions (OECD, 2019[11]), as well as for any market
where consumer demand requires a stream of innovative products, as for example in entertainment
production markets that require new content and new design.
Theories of harm to potential competition
Potential competition can be harmed by the same firm conduct and agreements that can prevent, restrict
or distort actual competition. This can take multiple forms, some of which are discussed below.
It should also be noted that there might be many other features of a market that prevent, restrict or distort
potential as well as actual competition. For instance, anti-competitive regulations, such as those identified
by the OECD’s competition assessment toolkit, often raise barriers to entry that will predominantly damage
potential rather than actual competition (OECD, 2019[12]). Similarly, the use of competitive neutrality tools
to ensure equal treatment of all firms, including State-Owned Enterprises, helps to create a level playing
field that may not only strengthen the competitive constraints between the existing products on the market,
but will also allow potential constraints to emerge and thrive.
Mergers
Mergers or acquisitions could remove the future constraint that a potential rival producer of a substitute
product would have placed on the incumbent. These might in some cases result in the incumbent shutting
down the acquired product or the (development of) its own rival version of the acquired product (a killer
acquisition, (OECD, 2020[4])). However, in other cases the loss of potential competitive constraints between
the incumbent and the acquired products might simply slow innovation, reduce quality or increase price,
but without affecting the choice of product as a killer acquisition would.
Alternatively, mergers or acquisitions might vertically integrate the incumbent’s product with a future
supplier or distributor, creating the same possibility of vertical input or customer foreclosure that might
need to be investigated in the case of vertical integration between existing products. As in vertical mergers
between existing products this would require that one of the products has a significant degree of market
power that might be protected or extended using the up or downstream product. 3
Exclusionary Practices
Dominant incumbents may have the incentive and ability to exclude potential competitors. Such
exclusionary practices can take multiple forms, and may even involve (individual or systematic) merger
and acquisitions, on their own or in tandem with other exclusionary practices. Indeed, in some cases it
might be more profitable for an incumbent to purchase the potential rival instead of adopting more elaborate
anti-competitive strategies through for instance exclusivity agreements or through predatory pricing. In
these cases, the acquisition itself might constitute exclusionary conduct. This for instance is increasingly
the allegation made by US agencies in cases such as Visa/Plaid (Box 2), PacBio (Box 1),
Facebook/Whatsapp and Facebook/Instagram.
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 11 Box 1. Pacific Biosciences / Illumina - merger and monopolisation case Illumina, a leading biotechnology firm active in sequencing technology sought to acquire rival Pacific Biosciences (PacBio). Reports suggest that PacBio had a current market share of just 2-3%. On 2 January 2020, the parties announced that they had agreed to terminate their merger agreement, following US and UK opposition. The US FTC had alleged that Illumina had sought to “unlawfully maintain its monopoly in the U.S. market for next-generation DNA sequencing (NGS) systems by extinguishing PacBio as a nascent competitive threat”. In particular, along with claiming under section 7 of the Clayton Act that the deal will eliminate current and future competition between the two companies, the FTC also investigated under section 2 of the Sherman Act, which prohibits attempting to obtain or maintain a monopoly. The UK CMA considered that the merger would result in a substantial lessening of competition in the supply of NGS systems in the UK. It noted that Illumina had approximately 80% market share of NGS systems worldwide and 90% in the UK. Through analysis of internal documents and customer feedback, the CMA found that the parties saw each other as a considerable threat, that there was some substitutability between their products and that competition between the parties would increase in the future due to PacBio’s advancements. The CMA noted that in the highly concentrated market, other small players in the sector would not exert a competitive constraint on the merged entity. Sources: 1 FTC Complaint of 17 December 2019, In re Illumina, Inc., & Pacific Biosciences of California, Inc. (PacBio). 2 CMA, Anticipated acquisition by Illumina, Inc (Illumina) of Pacific Biosciences of California, Inc. (PacBio), Summary of Provisional findings. Box 2. Visa/Plaid – merger and monopolisation case In November 2020, VISA’s proposed USD 5.3 billion acquisition of Plaid was abandoned following the US Department of Justice’s challenge of the transaction based on potential competition concerns. The department said that Visa made an offer for Plaid worth more than 50-times the target’s annual revenue on “strategic, not financial grounds” to “protect” the credit card company’s debit business. Plaid’s technology provides an interface for fintech apps to connect to users' bank with consumer permission. More concretely, when a consumer signs up with a Plaid-supported fintech app and provides her bank log-in credentials, Plaid uses those credentials to access the consumer’s financial institution and obtain the consumer’s financial data, which it transmits back to the fintech app. While this technology does not compete directly with Visa, the Department considered that Plaid’s extensive connections with banks and consumers acquired through this technology would give Plaid a unique position to offer a pay-by-bank debit service (a form of online debit that uses a consumer’s online bank account credentials rather than debit card credentials, facilitating payments to merchants directly from the consumer’s bank account). The department reported that Plaid indeed had plans to build on the success of its current services by creating a pay-by-bank debit service. CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
12 DAF/COMP(2021)3
Since Plaid’s future development of its own pay-by-bank debit service directly threatens Visa’s online
debit business, where Visa held a market share of approximately 70%, the department argued that
Visa’s proposed acquisition of Plaid would eliminate that nascent competitive threat and unlawfully
maintain Visa’s monopoly power in the online debit market. Similar to the FTC in Pacific Biosciences
/ Illumina, the department therefore challenged the merger under both section 2 of the Sherman Act
and section 7 of the Clayton Act. Visa argued that potential competition theories of harm have for
decades been evaluated under Section 7 and “have found almost no traction in the courts”. In
contrast, under section 2 it is not necessary to show that the potential competitive constraint that
has been excluded was likely to have become a competitive constraint.
Source: USA v. Visa Inc. et al., Case number 3:20-cv-07810, in the US District Court for the Northern District of California.
It is also possible that an acquisition strategy (which individually might be pro- or anti-competitive) can be
complemented by an associated threat to exclude (see (Kamepalli, Rajan and Zingales, 2020[13]).4 For
instance, such a threat might be deployed in order to bring reluctant targets to the negotiation table, and
to reduce the price they obtain when they get there (thereby reducing the incentive to invest in start-ups).
Investors often see the prospect of a start-up being acquired as an incentive to invest, since this can be a
very profitable exit strategy for them.5 However, if the acquisition practices of an incumbent were to create
the perception that there is a kill-zone where no new companies will be allowed to grow and compete, this
could instead have the opposite effect of reducing investment and potential market entry.
Beyond this, there are numerous exclusionary practices that do not involve the acquisition of competitors.
For example, there is a risk for a nascent firm that a gatekeeping platform will copy its product and compete
against it on the merits (i.e. within the boundaries of competition law). For example, Amazon has introduced
a number of its private-label items that directly compete with the products of independent sellers on its
platform. Such ‘genericisation’ is usually welcomed as pro-competitive, for example when a patent expires.
However, when it comes as quickly as it can do in the cases of a gatekeeping platform, the static efficiency
of the outcome can mask the damage done to dynamic competition, particularly when appropriability of an
innovation6 is low.7
As with the possibility of kill zones created by merger practices discussed above, opportunistic conduct by
the platform (extracting information and using it to copy and profit at the expensive of an innovative
originator) may create an investment hold-up problem that reduces the incentive for nascent firms to
innovate or invest in obtaining efficiencies. Moreover, it creates downward pressure on any future
acquisition price (which discourages investment in the first place). Indeed (Kamepalli, Rajan and Zingales,
2020[13]) and data quoted by the (Stigler, 2019[14]) 8 are each consistent with such an effect on innovation
incentives, as is the testimony of some investors.9
Competition agencies have begun recently to examine these issues as potentially exclusionary behaviour
(e.g. the Apple app store10 and Amazon probes11 in the EU). However, several authors have concluded
that such concerns are best addressed through ex-ante regulation and codes of conduct to ensure that
such platforms behave fairly with their suppliers (e.g. Proposed Digital markets Act in the EU, (CMA,
2020[15]), (Stigler, 2019[14]) and (Wheeler et al., 2020[16])).
Anticompetitive Agreements
Anticompetitive agreements may also be used to restrict or eliminate potential competition constraints.
These have been repeatedly investigated in pharmaceutical markets where pay-for-delay cases such as
in the Actavis (US)12 and Lundbeck (EU)13 cases. These judgements have confirmed that there does not
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 13
need to be an actual competitive constraint that is lost in order for a collusive agreement to be identified.
It is sufficient that the parties agree to remove a future constraint. Notably the courts have identified such
agreements as anticompetitive even where this future constraint is subject to a great deal of uncertainty
(e.g. over possible patent infringement actions).
The key parameters to identify potential competition
We have distinguished potential competition from the actual competitive constraints imposed by potential
entrants. That said, in many cases, potential and actual competitive constraints might co-exist
simultaneously. For example, a start-up (or a firm in an adjacent market) might pose a threat that make
the incumbent take certain steps to mitigate this threat (these steps reflect the strength of the actual
constraint). For instance, it prices lower than it otherwise might, or produces a better quality product than
it would do, absent that threat. However, at the same time, the very same start-up (or firm in an adjacent
market), might also pose a much larger potential constraint, in that if it were to enter, it would spark a much
more significant impact on price and quality.
Losing an actual constraint from a potential entrant has an impact that is certain (100%), because that
actual constraint, which was presumably already benefiting consumers and observable in the data, is
removed. However, while it is certain, it is likely to be smaller than the potential constraint, at least to the
extent that the incumbent decides to take only proportionate steps to mitigate the risk (or perhaps that the
incumbent unintentionally underestimates or entirely misses the risk and is blindsided). In contrast, the
potential constraint is inevitably uncertain (for example it might be 30% likely to manifest), but the impact
of losing that constraint, if indeed it is lost, might be more significant. That is, if the constraint were to
materialise, the impact on price, quality and innovation would likely be much more significant than the
actual constraint, since the incumbent would recognise the new certainty of the threat and react more
dramatically.
A potentially even more important distinction is that between potential competition and simple speculation.
That is to say, what are the limits of potentiality? Certainly, as every promising young musician or athlete
knows, potential may materialise, it may exist but ultimately go unfulfilled, or it may be mistakenly and
overoptimistically identified in the first place. How do we distinguish between these categories?
In the following Sections, we will consider some key parameters that need to be assessed in order to
confirm the relevance of a potential competitive constraint:
the relevance of barriers to entry;
the likelihood and strength of potential competition; and
the timeframe in which potential competition could emerge.
CONCEPT OF POTENTIAL COMPETITION @ OECD 202114 DAF/COMP(2021)3
3 Assessing Barriers to Entry
One feature of a market that can influence the strength of a potential competitive constraint are barriers to
entry into the market. This Section briefly sets out the definitions and forms of barriers to entry and suggest
that, as a result of the ambiguous impact of barriers to entry, an assessment thereof could not provide as
such a clear answer in cases where a specific potential competitive constraint is allegedly lost.
Definitions and forms of barriers to entry
In general, barriers to entry refer to an impediment that makes it more difficult for a firm to enter a market.
However, there has long been debate over the definition of a barrier, and hence on what is, and what is
not a barrier to entry. For competition enforcement purposes, this debate about what should be labelled a
barrier to entry is often not informative. Instead, competition authorities can focus on the likelihood and
timing of entry by specific firms into a market (OECD, 2005[17]). However, to the extent that the likelihood
of entry by a specific firm, and hence the potential constraint posed by that specific firm is influenced by
the presence of cross cutting barriers to entry it is useful to explore the concept of a barrier to entry.
(Bain, 1956[18]) said that the conditions of entry should be: “evaluated roughly by the advantages of
established sellers in an industry over potential entrants, these advantages being reflected in the extent to
which established sellers can persistently raise their prices above a competitive level without attracting
new firms to enter the industry”.
This led him to identify that economies of scale can be a barrier to entry since they meant that a new entrant
would have to choose between entering at a small, and hence less efficient scale, or entering at an efficiently
large scale that risks flooding the market and depressing prices, making entry unprofitable. Notably, Bain did
not require that incumbents had not faced the same obstacle in order to categorise it as a barrier.
Nevertheless, it is possible that the incumbent will not have faced the same entry conditions. For instance, if
the incumbent enjoyed a first mover advantage, or if market growth had been faster at the time of its entry.
In contrast, (Stigler, 1968[19]) adopted a much more restrictive definition, arguing that a barrier to entry was:
“... a cost of producing (at some or every rate of output) ... which must be borne by a firm which seeks to
enter an industry but is not borne by firms already in the industry”. He therefore did not consider economies
of scale to be barriers to entry, since he assumed that incumbents faced the same economies when they
entered. Similarly, product differentiation, capital costs, advertising costs, patents, or cost advantages were
all not considered to be barriers. 14
Following Bain and Stigler, numerous variations on these definitions have been proposed over the years
(see (McAfee, Mialon and Williams, 2004[20])). The most useful however is that offered by (Carlton and
Perloff, 2005[21]) who defined an entry barrier as “Anything that prevents a firm from instantly creating a
new firm in a market”. This recognises that in the short run, there are almost always some barriers to
entry, and so the textbook version of perfect competition is just that, a hypothetical example that amongst
many other things, lacks any time-dimension.15
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 15 Under this broad definition of a barrier to entry, there are a host of potential barriers to entry. These include not only economies of scale and scope, sunk costs, product differentiation and cost asymmetries, but also network effects, behavioural biases and trade and regulatory barriers. In recent years, the examples of demand side externalities such as network and cross-platform network effects have been identified as an important barrier in some markets (see for example (Furman Report, 2019[22]), and (Stigler, 2019[14])). Where these types of effects are strong, they make entry difficult, even for more efficient rivals. This is because a more efficient entrant needs to persuade users, not only that their product is better, but that it is sufficiently better to outweigh the loss of value that users would incur by switching to a network that is not interoperable with the mass of users on the incumbent platform. The difficulty in recruiting users when adoption is low means that entrants face an S-shaped adoption curve, and that incumbents, whatever their quality or efficiency, benefit from users valuing their product largely on the basis of who else uses the same product. The fact that value depends on other users, means that in the absence of interoperability, each user recruited by a platform will raise its rivals’ recruitment costs and soften competitive constraints. This creates a coordination problem that can mean that persistent bad equilibria can exist in which relatively poor quality or inefficient products continue to be in high demand despite users agreeing on their inadequacies. If the products are reasonably high quality, it then becomes extremely difficult for more innovative entrants to provide the additional value that is necessary to overcome these barriers and challenge them. The behavioural biases of consumers (such as aversion to search or risk, or heavy discounting of future aftermarket purchases) can also form a barrier to attracting demand, again even for more efficient, better quality products and services. In the past, a more simplistic economic analysis might have mistakenly assumed that decisions exhibiting these biases were the expression of consumer preferences, and hence not a barrier to entry, rather evidence of markets giving consumers what they want. However, modern economic analysis is live to the possibility that these biases can create barriers to entry that may require intervention in order to ensure a competitive market (CMA, 2017[23]). While it is true that one consumer’s bias might be another’s preference, a closer look at consumer decision-making is often sufficient to distinguish between markets in which choices largely reflect revealed preferences and those where they reflect biases. For instance, this analysis has helpfully identified that a lack of entry in some markets is not a result of anticompetitive behaviour by incumbents, nor anticompetitive regulation by government. In these cases, it has enabled competition policymakers to identify that effective competition requires interventions to improve the speed, reliability and security of switching services for mobile phone, banking accounts and electricity suppliers (OECD, 2018[24]). Box 3. What to do about cross-platform network effects and behavioural biases? In the course of undertaking market studies agencies may identify barriers to entry such as cross- platform network effects and behavioural biases as features of the market that restrict potential competition (without necessarily finding any fault with the behaviour of the firms that benefit from them). Alternatively, they might find in the context of an abuse of dominance investigation that a dominant firm is able to exploit and strengthen such barriers. For instance, purchasing a default position might create de facto exclusive dealing in the same way that a fidelity rebate can do, and may therefore exclude potential rival’s by raising their costs. Similarly a firm might withdraw or turn-off interoperability in order to undermine an entry strategy (see the shutdown of Vine and Twitter’s access to Facebook’s APIs (The Verge, 2018[25]), and more generally (OECD, 2021[26]). CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
16 DAF/COMP(2021)3
Where there is preliminary evidence to support an exclusionary theory of harm based on these types
of barrier to entry, agencies can, and have, launched investigations of the conduct of the dominant firm.
Since the excluded rival is a potential entrant, and entry is highly time-sensitive, these are likely to
require interim measures in order to have any hope of being effective. However, whether the barrier is
in some sense a natural feature of the market that is identified through a market study, or an
endogenous barrier that the incumbent is using to exclude, the problem of how to remedy the issue
remains. Here the most promising answers are of a regulatory nature and so these feature heavily in
the case being made for ex-ante pro-competitive regulation by Furman, Cremer, Stigler and others. For
example, interoperability that requires that potential rivals have access to standardised open APIs (see
(OECD, 2021[26]) and (OECD, 2020[27])).
Barriers to entry can also be regulatory or trade-related. These might for instance include tariff and non-
tariff trade barriers, such as subsidies, advantageous tax rates for certain producers, or weak employment
regulations (e.g. regulatory exemptions for ride-hailing platforms).
There can also be regulatory barriers of the type identified by the OECD’s competition assessment toolkit
(OECD, 2019[12]). These can include regulations requiring licensing. However, as with other barriers to
entry these might well be welfare enhancing. For example, standards for product labelling, either on a
products origins, or on its contents, might improve consumers ability to make informed decisions while
restricting entry into certain markets.
The expectation of future barriers to exit may also form a barrier to entry to the extent that they are factored
into a firm’s decision to enter. For example, the expected cost of redundancy payments and site clean-up
might discourage a firm in some cases. Similarly, the lack of an effective bankruptcy regime might increase
the risk of failure and hence discourage entry in the first place (OECD, 2019[28]).
Finally, barriers need not be absolute, in the sense that they may simply delay rather than prevent entry.
Such delays can have lasting effects as Mark Zuckerberg noted when considering Facebook’s acquisition
of Instagram: “There are network effects around social products and a finite number of different social
mechanics to invent. Once someone wins at a specific mechanic, it’s difficult for others to supplant them
without doing something different. […] One way of looking at this is that what we’re really buying is time.
Even if some new competitors springs up, buying Instagram, Path, Foursquare, etc now will give us a year
or more to integrate their dynamics before anyone can get close to their scale again. Within that time, if we
incorporate the social mechanics they were using, those new products won’t get much traction since we’ll
already have their mechanics deployed at scale. (US House of Representatives, 2020[29])”
What impact do barriers to entry have on potential competition?
Low barriers to entry mean that more entrants will have the opportunity to enter and compete, and make it
easier for those potential entrants to enter. We can consider the effect this has on both actual competitive
constraints, and potential competitive constraints, as well as on the overall constraints, and the importance
of a specific constraint.
Barriers to entry on overall constraints
The actual constraint posed by potential entrants as a group should be larger when barriers to entry are
low because there is a higher likelihood that one or more of the many possible entrants will enter. In this
case, a pre-emptive strategy taken by the incumbent against possible entrants would need to be significant.
However, even if barriers to entry are low, such the pre-emptive response may well be very small or
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 17
non-existent because the incumbent may choose to hold back its mitigation strategy until it is triggered by
actual entry, at which point it might then cut price to compete. In these cases, the actual constraint posed
by potential entrants as a group are likely to be small regardless of the strength of the barrier to entry.
Meanwhile, the potential competitive constraint that would materialise in the event that entry occurs will
not necessarily be any different when barriers to entry are low. However, since the probability of entry
increases as barriers fall, there will be less uncertainty as to whether that potential constraint will
materialise. Therefore, the likelihood and weighting attached to a group of potential competitive constraints
by an agency will be larger when barriers to entry are low.
Barriers to entry on specific constraints
However, what matters in mergers and exclusionary cases is a specific assessment of the prospects of
entry for an individual firm, while an analysis of cross-cutting barriers to entry discussed above might be
useful within the context of a market study,
The impact of high barriers
An analysis of the specific prospects of entry by an individual firm would consider the specific
characteristics and circumstances of a specific firm. It might therefore find that while there are generally
high barriers to entry, there is nevertheless a realistic prospect that these could be overcome by the firm
in question.
This matters because for an individual firm’s constraint on an incumbent to have a more significant effect
on competition, this would require that other potential third party entrants are less likely to pose a constraint.
Therefore, by reducing the threat of third party entry, higher barriers to entry will tend to increase the
substantiality of a constraint posed by an individual rival that does have a viable route to entry (a way over
the barrier). This is true of both potential and actual constraints posed by potential entrants.
Therefore, it does not follow from the above that we would worry more about a loss of potential competition
when barriers to entry are low and our confidence in the existence and strength of that potential constraints
is high. Rather, it is precisely when barriers are high and potential competitive threats are rare and perhaps
uncertain that we would worry about losing those rare threats that do have the potential to enter.
Identifying the existence of generally high barriers to entry which make potential entry less certain is
therefore likely to help support a theory of harm that acquisition, exclusion or collusion with a specific firm
is likely to restrict potential competition. However, this does rely on there being evidence that despite the
high barriers to entry, the acquired or excluded firm nevertheless has a credible specific route through that
barrier. This might, for example, be clearest where the entrant posed a threat based on a disruptive
innovation, or a maverick business model, or access to specific intellectual or physical property that others
might lack. In addition, the fact that the entrant has entered into similar markets in the past could also
constitute such evidence. The relative specificity of its ability to clear a high entry hurdle might also be
demonstrated if it has already entered while others either have failed, or have yet to try.
It is also possible that the firm that is most capable of overcoming otherwise high barriers to entry is a third
party that might mitigate the effects of a merger or exclusion. In such cases, the strength of barriers to
entry may provide interesting background, but a specific assessment of the prospects of entry for that
individual firm will again be necessary.
CONCEPT OF POTENTIAL COMPETITION @ OECD 202118 DAF/COMP(2021)3
The impact of low barriers
Should we therefore instead be relaxed about a loss of potential competition if barriers to entry are low?
Unfortunately, this is not straightforward either.
Lower barriers should, as noted, increase the likelihood that a specific potential competitive constraint
materialises, and hence increase the weight that agencies can place upon the loss of that specific potential
constraint. However, as also noted above, while the likelihood of the specific constraint arising might
increase, the likelihood of other constraints arising may also increase and make the loss of the particular
constraint less relevant. In order for the greater likelihood of a specific potential constraint to be relevant,
there would therefore need to be clarity on why further entry was not likely to follow (given the low barriers
that would do little to prevent it).
A theory of harm would therefore need to coherently explain that while further entry would not be prevented
(since barriers are low) by the incumbent’s conduct, it would not be expected to exert the same strong
constraint once in the market as the acquired, excluded or co-opted entrant would have. An asymmetry
between potential entrants would therefore need to be identified. This might, for instance, include
differences in their efficiency, their assets, or their product differentiation and hence the substitutability of
the products of the different potential entrants, or simply a strong first mover advantage.
Furthermore, low barriers to entry could also be consistent with a theory of harm in which an incumbent
was alleged to be engaged in a sustained attempt to prevent entry (e.g. by acquiring or excluding or forming
agreements with a series of potential rivals in order to build artificial barriers to entry). While Selten’s (1978)
chainstore paradox suggests this to be unlikely (Selten, 1978[30]), this, as Selten recognised, rests on
assumptions of perfect information and simultaneous, rather than sequential, entry, each of which are
unlikely to hold in practice. In such cases, the conduct itself seeks to create a higher barrier to entry in
order to substitute for the lack of exogenous barriers to entry. Indeed, the post-Chicago school of thought
has now identified numerous theories of harm under which there is an ability and incentive to exclude. The
possibility of such strategies might therefore deter highly rational potential entrants, and perhaps many
more potential entrants with more risk averse investors.
Finally, barriers might be endogenous in that they are determined by the behaviour of firms within the
market, or exogenous and therefore unaffected by that behaviour. Low exogenous barriers to entry might
therefore co-exist with high endogenous barriers. Therefore, Sutton’s point that competition in markets with
endogenous sunk costs can deter entry without any anticompetitive behaviour might still apply in markets
with low exogenous barriers. Endogenous barriers to entry are therefore one of the important features of
a market which can lead to the market working poorly for consumers, and which could therefore be
amenable to intervention via a market investigation, rather than through antitrust enforcement. For
instance, the responsiveness of demand to branding in certain markets can create the incentives that lead
to the creation of endogenous barriers to entry.
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 19
Summary
As a result of the ambiguous impact of barriers to entry, the height of such barriers does not
provide a reliable bright line safe harbour in cases where a specific potential competitive
constraint is lost. For instance, the height of a barrier to entry matters much less in cases
where the barrier applies to all, or whether it can be, is likely to be, or has already been,
overcome by specific firms.
This means that the existence of barriers to entry is not determinative, but rather one of the
elements that agencies will rely on to assess the likelihood of entry in the context of the
alleged theory of harm. As the European Commission has noted: “Rather than focusing on
whether "entry barriers" exist according to some definition, competition authorities should
explain how the industry will behave over the next several years and how rapidly and to
what extent entry could enhance competition. This means competition authorities should
assess the likelihood of entry, not whether entry barriers are high or low in any given case.
This implies that factors such as economies of scale, product differentiation or access to
scarce resources may all be entry barriers if their presence implies entry will be unprofitable
and thus unlikely”.
Simply measuring the height of barriers to entry will therefore rarely be an informative exercise
for competition authorities, which perhaps explains why such an exercise is rarely ever carried
out. We therefore turn in Sections 4 and 5 to the assessment of the likelihood and strength of
entry, and the timeframe within which it might occur.
CONCEPT OF POTENTIAL COMPETITION @ OECD 202120 DAF/COMP(2021)3
4 Assessing the Likelihood and
Strength of Potential Competition
If an assessment of barriers to entry provides few answers, and an entrant-specific inquiry is instead
required, then how do we assess the likelihood of a potential competitive constraint emerging? How do we
assess how strong that constraint would be? And how likely and how strong would a potential competitive
constraint need to be in order for it to be relevant for a decision?
This last question is a matter of thresholds, while the first questions focuses on the methodology for the
assessment of potential competition. We begin with the question of thresholds.
How likely and how strong do the thresholds for potential constraint need to be?
The different options that can be used as thresholds are easy to identify when we know probabilities and
the potential outcomes. Where we know these, as in the case of a coin toss (50% heads, 50% tails) or a
lottery ticket, they are risks (Knight, 1921[31]). When dealing with markets however, competition agencies
do not know the probabilities, nor the possible outcomes. They therefore have to assess these
uncertainties by estimating the shape or form of a specific entrant, the impact that it might have, and then
the likelihood of that combination occurring.
Having made an assessment, agencies compare this against the relevant thresholds and decision rules.
There might however, be a case for using different thresholds for potential competition from those that are
used when the concern is over the possible loss of an actual constraint. For example, the unobservable
nature of a potential constraint introduces considerable and inherent uncertainty. Moreover, it is not
straightforward to conclude that this inherent uncertainty should be treated in the same way as uncertainty
that arises from the possible mismeasurement of an actual competitive constraint. We explore this further
in the context of mergers and antitrust cases.
Mergers
The logical and economic approach would be to treat the potential constraint as equivalent to a certainty
(an actual constraint) by calculating an expected value for the constraint (likelihood multiplied by
magnitude). For example, the loss of a potential constraint that is expected to be significant, could then be
equivalent to an actual constraint that is significant. This for example was the proposal of the UK’s Furman
Review for a ‘balance of harms’ test which has also received support from many others (Furman Report,
2019[22]).16
Traditionally, however, merger control has not been based on this economic approach. Instead, we only
worry about the loss of entrants that were likely to enter, and, of those entrants that were likely to enter,
we only worry about those that would then apply a significant constraint.
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021 21 For example, Wu & Hemphill note that in the US mergers can be prohibited only where the competitor ‘probably’ would have entered the market and its entry would have had pro-competitive effects.17 They argue that, under this rule, the acquisition of a nascent competitor is nearly impossible to challenge, given the difficulty in establishing the “but-for” counterfactual world with sufficient precision and certainty. Indeed, the courts’ rejection of the US DoJ’s bid to block the recent Sabre/Farelogix merger (see Box 4) illustrated this difficulty. As such, they argue for a theory of harm based on ‘nascent competitors’ that reflects the innovation potential of such companies to be adopted. Box 4. Sabre/Farelogix merger The planned acquisition of Farelogix by Sabre was ultimately abandoned following close scrutiny of the deal by the US DoJ and the UK CMA. Sabre and Farelogix both provide technology solutions that facilitate airline bookings. Sabre is a major supplier of Global Distribution System (GDS) which facilitates transaction between airlines and travel agents. GDSs are two-sided platforms with sellers of travel services on the one side of the platform and airlines on the other. Farelogix does not operate a GDS but operates a separate technology, New Distribution Capability (NDC), which allows airlines to connect directly to travel agencies for bookings, without having recourse GDSs. The US DoJ challenged the transaction arguing that it would allow Sabre, the largest airline booking services provider in the US, to eliminate a disruptive competitor that has introduced a new technology to the travel industry and is poised to grow significantly. However, the DoJ’s attempt to block the transaction was denied by the court. The court held that Sabre’s GDSs, two-sided platform facilitating transactions between airlines and travel agencies, did not compete with Farelogix’s NDC, which only interacts with airlines and is not a two-sided platform, relying on the US Supreme Court’s American Express decision which held that “only other two-sided platforms can compete with a two-sided platform for transactions”. The court further noted that, even assuming that Sabre and Farelogix competed in the same market, the deal would not result in reduction in innovation. A few days after the US court’s decision, the UK CMA decided to block the transaction. Contrary to the findings of the US court, the UK CMA examined the potential competitive constraints and considered that GDSs compete with technologies that enable GDS bypass (e.g. NDC developed by Farelogix). While noting that Sabre and Farelogix were not close competitors, the UK CMA considered that Farelogix was a differentiated competitor to Sabre and the potential competition from Farelogix on Sabre’s GDS would be eliminated. It also noted that, absent the merger, Sabre could become a competitor to Farelogix’s NDC within the space of three to five years. Following the UK ruling the transaction was abandoned, and the US court judgement was vacated at the request of the DoJ who argued they would otherwise be unfairly denied the opportunity to appeal. Sources: 1. United States v. Sabre Corp., No. 1:19-cv-01548-LPS, 8 April 2020. 2. UK CMA, 9 April 2020, “Anticipated acquisition by Sabre Corporation of Farelogix Inc.: Final Report”. This same iterated framework is also applied to assess potential third party entry that might mitigate the anti-competitive effects of a merger or conduct. In that context, this means that potential competitive constraints need to firstly be likely to emerge, and then also likely to impose a significant constraint when CONCEPT OF POTENTIAL COMPETITION @ OECD 2021
22 DAF/COMP(2021)3
they do emerge. Only potential entrants that meet both thresholds would provide reassurance that a merger
that removed an actual constraint would in fact cause no harm.
Other thresholds for the merger test have been proposed, and many of them suggest the use of a different
threshold when assessing acquisitions of start-ups. For example, (Valletti, 2018[32]), (Crémer et al.,
2019[33]), (Motta and Peitz, 2020[34]), (Salop, 2020[35]) and (US House of Representatives, 2020[29])each
suggest that there should be a rebuttable presumption that acquisitions by dominant digital platforms are
anti-competitive unless the firms are able to demonstrate otherwise.
An alternative approach is to leave the initial burden of anticompetitive harm with the agency, but require
it only to show that there is a realistic prospect that an acquisition would be expected to reduce potential
competition.18 Where the agency manages to do so, this would then create a rebuttable presumption that
the merger would harm consumers, which the parties could then seek to overturn. This might be combined
with a shift to assessing the magnitude of harm as well as likelihood (as recommended by (Furman Report,
2019[22]) and (Caro de Sousa & Pike, 2020[36])). The (CMA, 2020[15]) have recently adopted a proposal
along these lines, and recommended that the UK legislate to change the evidentiary standard to one of a
realistic prospect of harm, at least in the case of acquisitions by firms with strategic market status.
Exclusionary Practices
The thresholds that are typically applied in mergers may not apply in anti-competitive exclusion cases. For
instance, in the US Microsoft case the courts were clear that it was not necessary to show that the potential
competitive constraint that had been excluded (Navigator and Java) were likely to have become a
competitive constraint. Instead it noted that “it would be inimical to the purpose of the Sherman Act to allow
monopolists free reign to squash nascent, albeit unproven competitors at will—particularly in industries
marked by rapid technological advance and frequent paradigm shifts.”19
This same approach is evident in the US DoJ’s challenging of the Visa/Plaid merger (see Box 2). This
merger was recently abandoned by the parties after the DoJ argued that the acquisition amounted to
unlawfully maintaining Visa’s monopoly over the online debit market in violation of Section 2 of the
Sherman Act (in addition to substantially lessening competition under Section 7 of the Clayton Act).
However, there is a strong case that despite this, the burden of proof in exclusionary cases remains too
high. For instance, (Gavil and Salop, 2020[37]) argue that using decision theory to set burden of proof would
lead to US courts setting a lower burden on the plaintiff in exclusionary conduct cases when the defendant
has substantial market power. This would reflect the fact that anticompetitive effects are more likely when
the defendant has substantial market power, and that there are asymmetric litigation incentives that tend
to produce false negatives.
This recognition of the need to shift burdens in the case of dominant firms is also evident in Europe. For
example, in Germany following the revisions to the Competition Law, the new Section 19a specifies that
in the case of undertakings with paramount significance for competition across markets, the burden of
proof in abuse cases now lies with the firms rather than the Bundeskartellamt. Similarly, the EU’s Digital
Markets Act specifies in articles 5 and 6 a series of obligations that will pre-empt the need to assess the
exclusionary or exploitative nature of conduct by digital gatekeepers.
Anticompetitive Agreements
How likely and how strong do the thresholds for potential constraint need to be in collusion cases? The
issue of potential competition has been relevant when assessing collusive agreements, for instance, in the
pharmaceutical industry. In some of these cases, entry has yet to materialise, such as in cases where
incumbents agree to pay-for-delay, and the generics manufacturers are, at the time of the agreement, not
CONCEPT OF POTENTIAL COMPETITION @ OECD 2021You can also read
